Explore Sequence Diversity

This page contains the contents for a jupyter notebook used to demonstrate using cazomevolve to explore the sequence diversity in the CAZy family PL20.

The original notebook can be found here and which can be used as a template.

Notebook

This notebook is for exploring protein sequence diversity in a set of CAZymes, e.g. a CAZy family.

Prior to using this notebook:

1. Build a local CAZyme db using cazy_webscraper

2. Retrieve the protein sequences for each CAZy family of interest using the cazomevole subcommand get_fam_seqs

3. Run all-vs-all analysis using BLAST or DIAMOND using the cazomevolve subcommands run_fam_blast and run_fam_diamond, respectively

This notebook takes as input the output from BLASTP+/DIAMOND and visualises the data.

Feel free to use this notebook as a template to perform further analyses.

Imports

import seaborn as sns

from cazomevolve.seq_diversity.explore.cazy import get_cazy_proteins, get_cazy_db_prots
from cazomevolve.seq_diversity.explore.parse import load_data, remove_redunant_prots
from cazomevolve.seq_diversity.explore.plot import plot_clustermap, plot_heatmap_of_clustermap

Constants

Define proteins of interest, e.g. proteins to be explored in the lab. These will be highlighed on the resulting clustermaps and heatmaps.

The dict uses the group name (e.g. a CAZy family) as the key, and is valued by a list of the NCBI protein version accessions.

CANDIDATES = {
    'PL20': ['EAA65331.1', 'UKZ86141.1', 'WAO96149.1']
}

Get ‘characterised’ proteins from CAZy

To retrieve proteins listed in the ‘characterised’ or ‘structure’ tables in CAZy, using the get_cazy_db_prots function.

We store proteins listed in the characterised table in the variable characterised_prots, and proteins in the structure table are listed in structure_prots.

characterised_prots = {}  # {fam: [prot acc]}
characterised_prots['PL20'] = get_cazy_db_prots('PL20', characterised=True)
characterised_prots

Parsing characterised table for PL20: 100%|██████████| 7/7 [00:00<00:00, 5420.00it/s]

{'PL20': ['BAG80639.1', 'OSZ99920.1', 'OTA07247.1']}

structured_prots = {}  # {fam: [prot acc]}
structured_prots['PL20'] = get_cazy_db_prots('PL20', structured=True)
structured_prots

Parsing structured table for PL20: 100%|██████████| 7/7 [00:00<00:00, 9031.11it/s]

{'PL20': ['BAG80639.1']}

Family analysis

Here is some example code for running the analysis for CAZy family PL20.

# load data
pl20_df = load_data('../../docs/examples/pl20/pl20_blastp', 'PL20')
pl20_df.head(5)

Identifying IPGs:   0%|          | 0/110 [00:00<?, ?it/s]

Identified 8 groups of identical proteins

	qseqid	sseqid	qlen	slen	length	pident	evalue	bitscore	BSR	qcov	scov
0	EAQ84327.1	EAQ84327.1	250	250	250	100.000	0.000000e+00	520.0	2.080	1.000	1.000000
1	EAQ84327.1	AEO59473.1	250	250	250	79.600	6.520000e-157	428.0	1.712	1.000	1.000000
2	EAQ84327.1	EDP51830.1	250	292	254	57.874	4.100000e-109	309.0	1.236	1.016	0.869863
3	EAQ84327.1	QSQ14196.1	250	255	251	53.785	3.570000e-101	287.0	1.148	1.004	0.984314
4	EAQ84327.1	QSQ27087.1	250	255	247	54.251	3.500000e-100	285.0	1.140	0.988	0.968627

# build clustermap of BLAST Score Ratio
pl20_bsr_plt = plot_clustermap(pl20_df, 'PL20', 'BSR', fig_size=(100, 100), save_fig='pl20_clustermap.png')
pl20_bsr_plt

<seaborn.matrix.ClusterGrid at 0x7fbb8c265460>

# plot a clustermap of only the candidates and functionally/structurally characterised proteins
# that is also annotated to differentiate, candidates and functionally/structurally characterised proteins
pl20_char_bsr_plt = plot_clustermap(
    pl20_df,
    'PL20',
    'BSR',
    fig_size=(7, 7),
    char_only=True,
    annotate=True,
    candidates=CANDIDATES, structured_prots=structured_prots, characterised_prots=characterised_prots,
)
pl20_char_bsr_plt

<seaborn.matrix.ClusterGrid at 0x7fbb7b99f520>

Then plot the precentage identity and query coverage for the candidate and functionally/structurally characterised proteins, plotting the proteins on the heatmaps in the same order as they appear in the clustermap.

print('PL20 percentage identity, colour scheme blue')
plot_heatmap_of_clustermap(
    pl20_char_bsr_plt,
    pl20_df,
    'PL20',
    'pident',
    fig_size=(7, 7),
    colour_scheme=sns.color_palette("Blues", as_cmap=True),
    candidates=CANDIDATES, structured_prots=structured_prots, characterised_prots=characterised_prots,
)

PL20 percentage identity, colour scheme blue

print('PL20 query coverage, colour scheme purple')
plot_heatmap_of_clustermap(
    pl20_char_bsr_plt,
    pl20_df,
    'PL20',
    'qcov',
    fig_size=(7, 7),
    colour_scheme=sns.color_palette("Purples", as_cmap=True),
    candidates=CANDIDATES, structured_prots=structured_prots, characterised_prots=characterised_prots,
)

PL20 query coverage, colour scheme purple